Probiotic Saccharomyces boulardii CNCM I-745 prevents outbreak-associated Clostridium difficile-associated cecal inflammation in hamsters

By utilizing a well-established hamster model of CDI we show that oral gavage of Saccharomyces boulardii CNCM I-745 (S.b) effectively prevented cecal tissue damage, NF-κB phosphorylation, and TNFα expression in hamsters infected with hypervirulent C. difficile strains. S.b culture supernatants also prevented toxin-mediated cellular actin disruption induced by conditioned medium from cultures of hypervirulent C. difficile strains. Thus S.b may reduce the risk of infection of several outbreak-associated C. difficile strains.

Abstract

C. difficile infection (CDI) is a common debilitating nosocomial infection associated with high mortality. Several CDI outbreaks have been attributed to ribotypes 027, 017, and 078. Clinical and experimental evidence indicates that the nonpathogenic yeast Saccharomyces boulardii CNCM I-745 (S.b) is effective for the prevention of CDI. However, there is no current evidence suggesting this probiotic can protect from CDI caused by outbreak-associated strains. We used established hamster models infected with outbreak-associated C. difficile strains to determine whether oral administration of live or heat-inactivated S.b can prevent cecal tissue damage and inflammation. Hamsters infected with C. difficile strain VPI10463 (ribotype 087) and outbreak-associated strains ribotype 017, 027, and 078 developed severe cecal inflammation with mucosal damage, neutrophil infiltration, edema, increased NF-κB phosphorylation, and increased proinflammatory cytokine TNFα protein expression. Oral gavage of live, but not heated, S.b starting 5 days before C. difficile infection significantly reduced cecal tissue damage, NF-κB phosphorylation, and TNFα protein expression caused by infection with all strains. Moreover, S.b-conditioned medium reduced cell rounding caused by filtered supernatants from all C. difficile strains. S.b-conditioned medium also inhibited toxin A- and B-mediated actin cytoskeleton disruption. S.b is effective in preventing C. difficile infection by outbreak-associated via inhibition of the cytotoxic effects of C. difficile toxins.

NEW & NOTEWORTHY

By utilizing a well-established hamster model of CDI we show that oral gavage of Saccharomyces boulardii CNCM I-745 (S.b) effectively prevented cecal tissue damage, NF-κB phosphorylation, and TNFα expression in hamsters infected with hypervirulent C. difficile strains. S.b culture supernatants also prevented toxin-mediated cellular actin disruption induced by conditioned medium from cultures of hypervirulent C. difficile strains. Thus S.b may reduce the risk of infection of several outbreak-associated C. difficile strains.

clostridium difficile infection (CDI) is one of the most common infectious diarrheas in hospitals and long-term care facilities in the United States related primarily to the use of antibiotics (46). C. difficile is an anaerobic bacterium that produces two toxins — toxin A and toxin B — that mediate diarrhea, inflammation, and apoptosis of the mucosal epithelium in animals and humans (32). The effects of the toxin in target intestinal cells involve inactivation of the Rho family of GTPase, leading to cytoskeletal disorganization, epithelial cell apoptosis, and ultimately cell death (42, 54). C. difficile toxins also stimulate transcription of several proinflammatory genes, including tumor necrosis factor-α (TNFα) (19) and activate transcription factors and mitogen-activated protein kinases involved in their proinflammatory effects (18, 25). The mainstream CDI regimen includes the use of metronidazole, vancomycin, and fidaxomicin (55). However, many C. difficile-infected patients may also suffer from recurrent infections (13), while the recent epidemics that also involve new epidemic outbreak-associated strains (29) pose a major medical problem and epidemiological concern. These challenges have been met with a broad range of preventive approaches against CDI, including the use of probiotics, most notably Saccharomyces boulardii (S.b) alone or in combination with established antibiotic treatment (23, 24).

S.b is a nonpathogenic yeast that represents one of the well-studied probiotics against CDI in both experimental and clinical settings (24, 40). Randomized double-blind placebo-controlled clinical trials have shown that S.b CNCM I-745 is an effective probiotic in the prophylaxis of antibiotic-associated diarrhea and the most effective probiotic of prophylaxis against CDI (36, 49). Studies also indicate that S.b can be used in combination with vancomycin as therapy for relapsing CDI (37, 50). The putative mechanisms involved in the effectiveness of S.b in CDI include effects directed against the microbiome, the host, as well as C. difficile and its toxins (4, 6, 21, 40, 43). However, most of the experimental work examining the effects of S.b in CDI models has been limited to non-outbreak-associated C. difficile strains, mostly with strain VPI10463. However, the increased incidence and severity of the CDI global outbreaks has been associated with the emergence of outbreak-associated strains (20). The exact mechanism of the hypervirulence of these emerging strains is not fully understood (20). Moreover, studies with the use of probiotics in emerging CDI at both the preclinical and clinical levels have been limited.

Based on these considerations, we examined the effectiveness of oral S.b administration in preventing morbidity, cecal mucosal histology, and cytokine expression in response to C. difficile infection due to the outbreak-associated strains ribotype 017, 027, and 078 in a well-established model of hamsters. These strains were chosen based on their impact in global outbreaks (51). Lastly, the effect of the S.b-conditioned medium in cytotoxicity caused by C. difficile filtered supernatants in mouse fibroblasts was also evaluated.

MATERIALS AND METHODS

C. difficile culture.

C. difficile strains VPI10463 (ATCC stock 43255), ribotype 027 (ATCC BAA-1805), ribotype 078 (ATCC BAA-1875), and ribotype 017 (ATCC 43598) were cultured in Difco cooked meat media (no. 226730 BD, Fisher Scientific, Canoga Park, CA) at 37°C in anaerobic conditions as previously reported (19).

Hamster model of C. difficile infection.

Six-week-old Golden Syrian hamsters (strain code 049) were purchased from Charles River Laboratories (San Diego, CA) and housed at the UCLA animal facility under standard conditions with a 12-h light period and a 12-h dark period per day at 25°C room temperature. Hamsters were housed in disposable plastic cages with HEPA-filtered air circulation, autoclaved bedding, standard animal chow, and sterile water ad libitum. Average body weight of hamsters is 120 g.

Hamsters (n = 5 per group) were orally administered with clindamycin (30 mg/kg in 200 μl dissolved in water, no. C5269, Sigma, St. Louis, MO) on day 1, followed by C. difficile (VPI10463, ribotype 017, 027, and 078, 10³ CFU in 100 μl) infection via oral gavage on day 5 as previously described (10). Some hamsters were orally fed with live or heated S.b (3 g/kg body wt in 200 μl dissolved in water) twice a day from day 1 to day 7. The 3 g/kg contains 3 × 10¹⁰ live yeast cells/kg, i.e., ∼3.6 × 10⁹ per hamster. A positive control group received vancomycin (50 mg/kg, dissolved in water) orally once a day from day 1 to day 7 while a negative control group received live and heat-treated S.b without C. difficile infection.

Lyophilized S.b CNCM I-745 powder was provided by Biocodex (Gentilly, France). Lyophilized S.b was dissolved in water to produce live culture for immediate oral administration. Heat-treated S.b was prepared by heating the live culture in a microwave oven for 20 s, similar to previously used approaches (2, 9). The temperature of the heated S.b culture reached 95°C. Heated S.b cultures were then centrifuged and filtered (0.22-μm filter) to generate cell-free supernatants. Heated S.b cell-free supernatants failed to prevent toxin A- and B-mediated cell rounding in 3T3-L1 cells. The details of cell rounding experiments are explained under Cell rounding experiments below.

Details of C. difficile infection, vancomycin treatment protocol, and S.b treatment protocol are described in Fig. 1A, 2A, and 3A. All animals were euthanized by carbon dioxide gas at day 7. Hamster cecal content, cecal tissue, and serum samples were obtained at day 7 for further analysis. Animal experiments were approved by the UCLA Animal Research Committee (protocol 2007-116).

Fig. 1.

S. boulardii reduced C. difficile VPI10463-induced cecal tissue damage in hamsters. A: experimental design of C. difficile infection and Saccharomyces boulardii CNCM I-745 (S.b; SB) treatment. B: H&E staining of cecal tissues. Infection of C. difficile VPI10463 caused severe cecal tissue damage that was reduced by oral gavage administration of live S.b from day 1 to day 7. Administration of heated S.b had no protective effect. C: H&E staining of cecal tissues from hamsters without C. difficile infection. Magnification ×100. Black bars indicate 200 μm. D: histology score. Live S.b significantly reduced histology score of C. difficile VPI10463-infected hamsters.

Fig. 2.

Vancomycin reduced C. difficile VPI10463-induced cecal tissue damages in hamsters. A: experimental design of C. difficile infection and vancomycin treatment. B: H&E staining of cecal tissues. Infection of C. difficile VPI10463 caused severe cecal tissue damage that was reduced by oral gavage administration of vancomycin from day 1 to day 7. Magnification ×100. Black bars indicate 200 μm. C: histology score. Vancomycin significantly reduced histology score of C. difficile VPI10463-infected hamsters.

Fig. 3.

S. boulardii reduced cecal tissue damage in hamsters infected by outbreak-associated C. difficile ribotype 017, 027, and 078. A: experimental plan of outbreak-associated C. difficile infection and S.b treatment. B: H&E staining of cecal tissues. Infection of C. difficile ribotype 017, 027, and 078 caused severe cecal tissue damage that was reduced by oral gavage administration of live S.b from day 1 to day 7. Magnification ×100. Black bars indicate 200 μm. C: histology score. Live S.b significantly reduced histology score of C. difficile ribotype 017-, 027-, and 078-infected hamsters, respectively. Administration of heated S.b had no protective effect.

Histology scoring.

Cecal tissues in hematoxylin and eosin (H&E) staining were used for histology scoring. The severity of enteritis and colitis was graded using three parameters as previously published (41): 1) epithelial tissue damage; 2) hemorrhagic congestion and mucosal edema; 3) neutrophil infiltration. A score of 0–3 was assigned to each parameter. Total histology score was determined by the sum of all these three parameter scores (0–9). Four different locations per tissue section were observed. Each tissue section represented results from one hamster.

Cecal TNFα measurement.

The cecal levels of proinflammatory hamster TNFα was determined by ELISA kits (MBS2602630, mybiosource.com, San Diego, CA) according to the manufacturer's instructions.

Phosphorylated NF-κB immunohistochemistry.

Cecal tissues were fixed in 4% paraformaldehyde and embedded in paraffin. After incubation with blocking buffer, sections were incubated with a rabbit polyclonal anti-phosphorylated NF-κB p65 antibody (ab86299, Abcam, Cambridge, UK; 2 μg/ml dilution) overnight at 4°C. After washing, sections were incubated with donkey anti-rabbit IgG (19) and slides were stained with an ABC kit for color development (sc-2018, Santa Cruz, Dallas, TX). Images were analyzed with a Zeiss AX10 microscope. The histology core facility of the University of California Los Angeles provided assistance in H&E staining and immunohistochemistry experiments. We have verified that the NF-κB signal was specific as the tissue slides incubated with secondary antibody but without primary antibody had no detectable signal (data not shown). The quantitative difference of phosphorylated NF-κB signal was observed and scored: 0 = normal; 1 = mild; 2 = moderate; and 3 = strong. Four different locations per tissue section were observed and scored. Each tissue section represented results from one hamster.

Cell rounding experiments.

Mouse 3T3-L1 fibroblasts were cultured in DMEM with 10% fetal bovine serum (FBS) and 1% penicillin/streptomycin (10⁵ cells/well) in six-well plates. Cells were grown in 2 ml of medium per well to around 70% confluence. Cells were serum starved overnight and then incubated with serum-free DMEM medium containing S.b-filtered supernatants for 4 h, followed by toxin A or B (0.1 μg/ml) for additional 4 h. The toxin A was purified from C. difficile VPI10463 as previously described (17, 45). After 4 h, microphotographs to observe cell rounding were taken in a blinded manner as we previously described (28). The quantitative difference of cell rounding was observed and scored: 0 = normal; 1 = mild cell round; 2 = moderate cell rounding; and 3 = severe cell rounding. Four different locations were scored per cell culture well.

Some of the cells were treated with cell-free supernatant of C. difficile-conditioned medium instead of toxin A or B. C. difficile (VPI10463, ribotype 017, 027, and 078) was cultured for 1 wk at 37°C in the anaerobic condition as mentioned above. C. difficile culture suspension was centrifuged at 4,000 rpm at 4°C for 15 min. The supernatant was transferred to a new tube and then filtered through a 0.22-μm filter to remove cells. The cell-free supernatant was added to the 3T3-L1 fibroblast culture (1:10) and further incubated for 4 h.

To prepare S.b-filtered supernatants, lyophilized S.b powder was dissolved in DMEM media (100 mg/ml, i.e., 10%) overnight in 37°C with shaking as previously described (6). The culture was then centrifuged at 4,000 rpm at 4°C, followed by a filtration through a 0.22-μm filter to remove the yeast cells. The filtered supernatants were then used to treat the serum-starved cells. The control group was incubated with serum-free DMEM without S.b-filtered supernatant.

Actin cytoskeleton staining.

Culture conditions of cells in the ActinGreen staining studies were the same as the cell rounding experiments. Two drops of ActinGreen reagent (R37110, ThermoFisher, Canoga Park, CA) were added to the culture (1 ml) during the last hour of the experiments. At the end of the incubation period, one drop of DAPI solution (Sc-24941, Santa Cruz) was added to the cell cultures and a coverslip was placed. Stained cells were observed under a confocal microscope: green = actin; blue = nuclei.

PCR analysis of C. difficile toxin A and B genes in cecal content.

To detect presence of C. difficile toxin A (tcdA) and toxin B (tcdB) genes in cecal contents, fresh cecal contents were collected into Eppendorf tubes and snapped frozen in dry ice, followed by storage in −80°C. We used 200 mg of cecal content per hamster to prepare bacterial DNA using QIAamp DNA Stool Mini Kit (no. 51504, Qiagen, Valencia, CA). The DNA concentration was measured by a Nanodrop machine and was then normalized to 100 ng per PCR reaction with water. The PCR detection of tcdA and tcdB was performed in separate reactions as previously described (34). All sample reactions were performed in duplicate. After 40 cycles of PCR amplification, the Ct values were used to evaluate the presence or absence of tcdA and tcdB genes. High Ct value (∼35–40) or absence of Ct value indicates the absence of tcdA and tcdB in cecal content. Low Ct value below 35 indicates the presence of tcdA and tcdB in cecal content.

Statistical analysis.

Quantitative results were expressed with error bars as means ± SE. Results were analyzed using Prism professional statistics software program (GraphPad, San Diego, CA). We used Mann-Whitney U two-tailed test for intergroup comparison unless otherwise specified. For all animal experiments, n = 5 hamsters per group.

RESULTS

Live, but not heated, S. boulardii significantly inhibited C. difficile strain VPI10463-mediated cecal inflammation in hamsters.

As shown by a previous report, live S.b is effective in protecting from C. difficile infection in hamsters (5). To establish this model in our laboratory, we administered clindamycin via oral gavage 5 days before C. difficile inoculation as suggested by a previous report (10). Hamsters were then infected with C. difficile strain VPI10463 as described in Fig. 1A. After 48 h, we collected cecal content, cecal tissues, and serum samples for further analyses.

We observed that C. difficile-mediated intestinal inflammation was evident primarily in the cecum (Fig. 1B). Tissue damage was characteristic of mild mucosal erosion/ulceration, severe neutrophil infiltration, and moderate mucosal edema. Images of C. difficile-infected hamsters are shown for evaluation (Fig. 1B). These quantitative changes were assessed by a histological score previously used by us (19, 41). Oral administration of live S.b improved C. difficile-mediated cecal tissue damage as shown by the reduced histology score (Fig. 1, B and D), consistent with a previous report (5). However, this improvement was not observed in hamsters fed with the same amount of heated S.b. This finding indicates that the anti-inflammatory effect of S.b was mediated by live S.b only.

To determine whether S.b alone alters cecal mucosal structure in hamsters, we treated hamsters with live and heated S.b without C. difficile infection. We did not observe any change in cecal tissue histology after live or heated S.b administration for 7 days (Fig. 1, C and D).

In the clinical setting, C. difficile infection is commonly treated with antibiotics such as metronidazole, vancomycin, or fidaxomicin (46). Similar to a previous study using a hamster model (27), vancomycin treatment altered the severity of CDI, reflected by reduced cecal tissue damage and the corresponding histological score in infected hamsters (Fig. 2, B and C).

Live, but not heated, S. boulardii significantly inhibited outbreak-associated C. difficile strain-mediated cecal inflammation in hamsters.

The incidence and severity of C. difficile outbreaks have been increasing with the emergence of outbreak-associated strains (20). We next infected hamsters with C. difficile strains ribotype 017, 027, 078, previously associated with several outbreaks (51), using the experimental protocol described in Fig. 3A.

Our results show that ribotype 017-, 027-, and 078-mediated cecal tissue damage was similar to what VPI10463 did with comparable histology scores (Fig. 3, B and C). Live, but not heated, S.b administration significantly reduced cecal tissue damage and corresponding histology score in hamsters infected with these outbreak-associated C. difficile strains (Fig. 3, B and C).

Live S. boulardii, but not heated S. boulardii, reduced outbreak-associated C. difficile strain-induced cecal proinflammatory TNFα protein expression and NF-κB phosphorylation.

As shown by our previous report (19), C. difficile infection results in an increased expression of the proinflammatory cytokine TNFα in the intestine. In our hamster model, infection with the common C. difficile strain VPI10463 leads to increased cecal TNFα protein expression, compared with cecal tissues without C. difficile infection (Fig. 4A). Oral vancomycin administration significantly reduced cecal TNFα expression in C. difficile-infected hamsters, suggesting successful suppression of C. difficile infection (Fig. 4B). Live, but not heated, S.b significantly reduced cecal TNFα expression in all C. difficile-infected hamsters by ∼30% (Fig. 4, A and C), while live or heated S.b did not affect basal level of cecal TNFα expression in hamsters without C. difficile infection (Fig. 4A).

Fig. 4.

S. boulardii reduced cecal TNFα protein expression induced by multiple C. difficile strains. A: cecal TNFα protein expression. Live S.b but not heated S.b reduced cecal TNFα protein expression in C. difficile strain VPI10463-infected hamsters. Cecal TNFα protein expression was low in normal hamsters without C. difficile infection and was not affected by live and heated S.b administration. B: vancomycin administration reduced cecal TNFα protein expression in C. difficile VPI10463-infected hamsters. C: live S.b but not heated S.b reduced cecal TNFα protein expression in C. difficile-infected hamsters, ribotypes 017, 027, and 078, respectively. Administration of heated S.b had no protective effect.

Also, our previous report showed that toxin A mediates TNFα expression via NF-κB activation (19). Colonic tissues from C. difficile-infected mice and ileal tissues of toxin A-treated mice had increased expression of phosphorylated NF-κB (19). Here we detected NF-κB phosphorylation in hamster cecum using immunohistochemistry. Infection with all C. difficile strains (VPI10463, ribotype 017, 027, and 078) led to a stronger NF-κB phosphorylation in the cecal mucosa than those of uninfected hamsters (shown in brown, Fig. 5, A–E). Live but not heated S.b substantially reduced NF-κB phosphorylation signal in the cecal mucosa of C. difficile-infected hamsters (Fig. 5, A–D). The basal level of phosphorylated NF-κB signal in the cecal mucosa of uninfected hamsters was very low (Fig. 5E). Quantitative assessment of phosphorylated NF-κB signal was shown in Fig. 5F.

Fig. 5.

S. boulardii reduced cecal phosphorylated NF-κB expression in hamsters infected with multiple C. difficile strains. A–D: phosphorylated NF-κB expression was heavily stained brown in the cecal mucosa of C. difficile-infected hamsters. Live S.b but not heated S.b reduced phosphorylated NF-κB expression in the cecal mucosa of infected hamsters, C. difficile strain VPI10463 and ribotypes 017, 027, and 078, respectively. E: cecal phosphorylated NF-κB expression in uninfected control hamsters. Magnification ×100. Black bars indicate 200 μm. F: quantification of phosphorylated NK-κB signal intensity.

C. difficile toxins A and B cause apoptosis in intestinal tissues (19). We used TUNEL assay to detect apoptosis in cecal tissues as previously described (19). Infection of hamsters with all C. difficile strains (VPI10463, ribotype 017, 027, and 078) led to a small number of apoptotic cells in the cecal mucosa as shown by the intense brown spots (Fig. 6, A–D). Live or heated S.b did not alter the number of apoptotic cells in the cecal mucosa of C. difficile-infected hamsters (Fig. 6, A–D). The apoptotic cells in the cecal mucosa of uninfected hamsters were almost undetectable (Fig. 6E).

Fig. 6.

S. boulardii did not alter apoptosis in cecal mucosa caused by multiple C. difficile strains. A: apoptosis is absent in uninfected control hamsters. B–E: apoptotic cells are stained in intense brown spots in the cecal mucosa of C. difficile-infected hamsters. TUNEL staining shows that live and heated S.b did not alter the mild apoptosis in the cecal mucosa of infected hamsters, C. difficile strain VPI10463 and ribotypes 017, 027, and 078, respectively. Magnification ×200. Black bars indicate 200 μm.

We detected whether S.b treatment affects the quantity of C. difficile in the cecum of hamsters using PCR. PCR results indicated the presence of tcdA in cecal contents of hamsters infected with VPI10463 and ribotype 027 and 078 (Fig. 7A). tcdA was undetectable in ribotype 017-infected hamsters, consistent with the lack of expression of the tcdA in ribotype 017 (Fig. 7A). tcdB was present in cecal contents of all C. difficile-infected hamsters (Fig. 7B). Live and heated S.b did not affect the quantity of tcdA and tcdB in the cecal content of C. difficile-infected hamsters (Fig. 7, A and B). This finding suggests that the protective effect of S.b does not depend on an antibacterial effect against C. difficile (Fig. 7C).

Fig. 7.

S. boulardii did not affect the presence of C. difficile bacteria in the cecum of the hamsters. A: Ct values of toxin A DNA in the hamster cecum detected by PCR. The high Ct value of uninfected control hamsters indicates the absence of C. difficile toxin A DNA in the cecum. The cecal content of hamsters infected with C. difficile VPI10463 and ribotype 027 and 078 had low Ct values that indicate the presence of C. difficile toxin A DNA. C. difficile toxin A DNA Ct values were not affected by live or heated S.b. Toxin A DNA was undetectable in ribotype 017-infected hamsters, as represented by high Ct values. B: Ct values of toxin B DNA in the hamster cecum detected by PCR. The high Ct value of uninfected control hamsters indicates the absence of C. difficile toxin B DNA in the cecum. The cecal content of hamsters infected with all C. difficile strains had low Ct values that indicate the presence of C. difficile toxin B DNA. C: presence of toxin A and toxin B in the cecal content of hamsters. All PCRs were performed with 100 ng cecal content DNA per well. All samples were assayed in duplicate wells.

S. boulardii-filtered supernatant prevented toxin A-mediated cell rounding and prevented C. difficile supernatant-mediated cell rounding.

Toxin A and B disrupt the actin cytoskeleton that leads to cell rounding and then cell death (56). To better understand the protective mechanism of S.b against toxins produced by C. difficile, we first examined whether cell-free S.b supernatant can prevent toxin A-associated cell rounding. Exposure of 3T3-L1 fibroblasts to purified toxin A for 4 h resulted in cell rounding (Fig. 8A). Consistent with a previous study (8), pretreatment of cells with S.b cell-free filtered supernatant (10%) partially prevented the cell rounding effect of toxin A. S.b-filtered supernatants alone without toxin A did not affect cell morphology of 3T3-L1 cells. These results indicate that S.b protects cells against toxin A-mediated cytoskeletal disruption in 3T3-L1 cells.

Fig. 8.

S. boulardii-filtered supernatant prevented cell rounding caused by purified toxins A and B. A: serum-starved 3T3-L1 preadipocytes were treated with DMEM or S.b-filtered supernatant (10%) in serum-free DMEM for 4 h, followed by PBS or C. difficile toxin A (0.1 μg/ml) for an additional 4 h. B: serum-starved 3T3-L1 preadipocytes were treated with DMEM or S.b-filtered supernatant (10%) in serum-free DMEM for 4 h, followed by PBS or C. difficile toxin B (0.1 μg/ml) for an additional 4 h. C: cell rounding scores. Exposure to toxin A and B caused the loss of the spindle shape of preadipocytes, but this change was prevented by coincubation with the S.b-filtered supernatant. We used t-tests for comparing between control group and toxin A/B-treated group since the cell rounding score of control groups were zero. Mann-Whitney test is not applicable in this scenario. The results are representative of 2 independent experiments.

On the other hand, toxin B also caused cell rounding in 3T3-L1 cells (Fig. 8B). Pretreatment of cells with cell-free S.b filtered supernatant (10%) partially prevented the cell rounding effect of toxin B (Fig. 8B). This finding may provide an explanation for the protective effect of S.b on C. difficile ribotype 017 infection since this strain produces only toxin B (11, 47). Quantitative assessment of cell rounding is shown in Fig. 8C.

To determine whether S.b protects cells from cell rounding due to outbreak-associated C. difficile, we first cultured C. difficile (toxin A+B+ VPI10463, ribotype 027, 078 and toxin A-B+ ribotype 017) for 1 wk and then prepared cell-free filtered supernatants. C. difficile supernatants caused cell rounding as observed with purified toxin A (Fig. 9, A–D). Pretreatment of cells with cell-free filtered S.b supernatant reduced cell rounding caused by C. difficile supernatant (Fig. 9, A–D). Quantitative assessment of cell rounding is shown in Fig. 9E.

Fig. 9.

S. boulardii-filtered supernatant prevented cell rounding caused the supernatant of multiple C. difficile strains. A–D: serum-starved 3T3-L1 preadipocytes were treated with DMEM or S.b-filtered supernatant (10%) in serum-free DMEM for 4 h, followed by filtered supernatant (10%) from the culture of C. difficile (strain VPI10463, ribotypes 017, 027, 078) for an additional 4 h. E: cell rounding scores. Exposure to C. difficile-conditioned supernatant caused the loss of the spindle shape of 3T3-L1 cells, but coincubation with the S.b-filtered supernatant partially prevented this change. The results are representative of 2 independent experiments.

The actin cytoskeleton is a known target of toxins A and B (14, 22). To visualize the distribution of intracellular actin fibers, we stained 3T3-L1 fibroblasts with the ActinGreen reagent. As shown in Fig. 10, toxins A and B disrupted the actin cytoskeleton network in these cells. Addition of S.b supernatants partially inhibited toxin A- and B-mediated actin network disruption, providing a mechanism for the reduction of toxin A- and B-associated cell rounding.

Fig. 10.

S. boulardii-filtered supernatant prevented actin disruption caused by purified toxins A and B. Serum-starved 3T3-L1 preadipocytes were treated with DMEM or S.b-filtered supernatant (10%) in serum-free DMEM for 4 h, followed by PBS or C. difficile toxin A or B (0.1 μg/ml) for an additional 4 h. The ActinGreen solution was added to the culture during the last hour of the experiment. Exposure to toxin A and B caused the disruption of the intracellular actin network, but this change was prevented by coincubation with the S.b-filtered supernatant. The results are representative of 2 independent experiments.

DISCUSSION

This study demonstrated that prophylactic oral administration of S.b to hamsters significantly prevented cecal tissue damage, TNFα protein expression, and NF-κB phosphorylation caused by three well-established C. difficile outbreak-associated strains. We also show that conditioned media from S.b. CNCM I-745 cultures substantially inhibited cell rounding caused by supernatants from outbreak-associated ribotype 017, 027, and 078 C. difficile strains. S.b supernatants also prevented toxin A- and B-mediated cell rounding and actin cytoskeleton network disruption, suggesting that this response is, at least in part, responsible for the in vivo protective effect seen in our study. Toxins A and B induce NF-κB phosphorylation that activates TNFα expression (19). TNFα may also cause cell death (33) and indirectly augments toxin A-mediated cytotoxicity (38). Thus the anti-NF-κB in vivo response afforded by S.b. shown here may represent an additional mechanism of action of this probiotic yeast.

Actin disruption also causes downstream NF-κB activation (30), which is linked to inflammatory cytokine expression. We had demonstrated that NF-κB activation mediates toxin A-induced TNFα expression in monocytes and macrophages (19). The S.b-mediated protection against toxin-mediated actin disruption may be pivotal in inhibiting cell death and inflammatory responses. S.b-conditioned medium also prevented the cell rounding cytotoxic effects of cell-free filtered supernatants from ribotype 017, 027, and 078, suggesting an effect on the cellular effects of C. difficile toxins produced by these strains. Most importantly, however, our results suggest S.b CNCM I-745 administration may be effective in preventing CDI caused by outbreak-associated C. difficile strains.

Multiple additional mechanisms can mediate the S.b. CNCM I-745 effect observed in our study. For example, a previous report demonstrated that S.b increases anti-toxin A IgA secretion in the small intestine (43), suggesting that small intestinal IgA may protect the host from mucosal damage during C. difficile infection. S.b also modifies the human gut microbiome by enhancing short-chain fatty acid-producing bacteria, including Lachnospiraceae and Ruminococcaceae (39). Short-chain fatty acids have known anti-inflammatory effects. S.b can also release a protease that digests toxin A and its receptor, thereby inhibiting the toxin effects (3, 4). It is not clear whether S.b can prevent CDI relapse or whether relapse will take place after withdrawal of S.b, but these issues worth further investigation in the future.

It is quite interesting that S.b had an inhibitory effect against all outbreak-associated C. difficile strains in vivo and in vitro, although the molecular and clinical characteristics of these strains are quite diverse. C. difficile ribotype 017 has been associated with several outbreaks CDI (11, 47) and is characterized by a toxin A-negative and toxin B-positive genotype (26). According to a mouse study, antibiotic treatment even promotes spore shedding of C. difficile ribotype 017 and subsequent animal-to-animal transmission (31). Since its tissue-damaging effect is thought to be primarily mediated by toxin B (26), the anti-cell rounding activity of S.b in our study indicates its protective effects against toxin B (Figs. 8 and 9).

Our results also show effective inhibition of C. difficile ribotype 027-induced intestinal inflammation mediated by S.b treatment. C. difficile ribotype 027 is one of the most serious outbreak-associated C. difficile strains responsible for multiple outbreaks in North America, Europe, and beyond (7, 53). It produces both toxin A and toxin B and an additional toxin, the CDT binary toxin (12). Although the mechanism of its hypervirulence is not fully elucidated, this strain is fluoroquinolone resistant (12) and appears to produce more toxin A and toxin B than any other C. difficile strains (35). The binary toxin also possesses cytotoxic activity, but its role in CDI is still not well understood (48). Based on these considerations, the in vivo and in vitro effects of S.b in C. difficile ribotype 027 may be directed against toxin A, toxin B, and/or the binary toxin. Interestingly, another probiotic, Lactobacillus acidophilus GP1B, inhibited C. difficile growth and transcriptional activation of C. difficile toxin-related genes in vitro (57).

We present here evidence for effective inhibition of C. difficile ribotype 078-associated cecitis mediated by S.b exposure. C. difficile ribotype 078 used to be prevalent among cows and pigs (1). Recent ribotype 078 outbreak data showed that it was similar to C. difficile ribotype 027 in severity, but C. difficile ribotype 078 tends to affect young people and causes community-associated outbreaks (16). Like C. difficile ribotype 027, ribotype 078 expresses toxin A, toxin B, as well as a binary toxin (44). The effects of S.b against CDI ribotype 078 seen in our results can be directed against these toxins.

Filtered supernatants of S.b cultures exert protective effects against C. difficile toxin A via inhibition of proinflammatory chemokine interleukin-8 expression in cultured human NCM460 colonocytes via an ERK-dependent mechanism (6). In a mouse ileal loop model, injection of S.b-filtered supernatants into the loops prevented toxin A-induced tissue damage, ERK activation, and chemokine KC expression (6). This finding is consistent with our results, indicating that S.b reduced cecal expression of the proinflammatory cytokine TNFα expression in C. difficile-infected hamsters in response to all outbreak-associated strains.

The overall protective effect of S.b against C. difficile infection in our studies is preventive rather than therapeutic. Once the infection started, S.b was ineffective in suppressing preexisting intestinal inflammation in hamsters (data not shown). It was necessary to start the S.b treatment before C. difficile infection and continue the treatment postinfection. This suggests that S.b requires some time to be established in the intestine to exert its protective effects. In our study, the three outbreak-associated strains are also hypervirulent strains and their enhanced toxicity was associated with a high mortality rate within a short period of time. The typical 10% S.b solution regimen in the drinking water used with regular C. difficile strains had no protective effect (data not shown) (52). We used the maximal dose of S.b within its solubility limit (3 g/kg, twice per day) to achieve successful inhibition of C. difficile-mediated cecal damage and inflammation.

In summary, we demonstrated that the preventive effect of S.b against C. difficile in several outbreak-associated C. difficile strains was similarly effective. This study is consistent with a recent meta-analysis study that demonstrated a statistically significant efficacy in the prevention of C. difficile diarrhea (15). We suggest that S.b CNCM I-745 pretreatment may reduce the risk of infection of several outbreak-associated C. difficile strains.

GRANTS

This work was supported by a grant from Biocodex Inc. NIH K01 (DK084256) and R03 (DK103964) grants (H. W. Koon). The Blinder Research Foundation for Crohn's Disease, NIH P30 DK 41301-26 (Animal Core), and the Eli and Edythe Broad Chair (C. Pothoulakis) also provided financial support. C. P. Kelly is supported by NIH grant R01 AI095256 and X. Chen by a Career Development Award from the Crohn's and Colitis Foundation of America.

DISCLOSURES

Biocodex Company sponsored this study.

AUTHOR CONTRIBUTIONS

H.W.K. and C.P. conception and design of research; H.W.K., B.S., C.X., C.C.M., D.H.-N.T., E.C.L., C.O., J.W., J.E.L., S.H., X.C., and C.P.K. performed experiments; H.W.K. analyzed data; H.W.K. interpreted results of experiments; H.W.K. prepared figures; H.W.K. drafted manuscript; H.W.K. and C.P. edited and revised manuscript; C.P. approved final version of manuscript.